Scott Winship on mobility in America

On net, it seems it is not going down and may be rising:

…consider what we know from previous studies of trends in intergenerational income mobility. The bulk of the existing research shows either that mobility has increased over the long run or that it has changed little in either direction. That includes both studies that I know of examining changes in upward mobility from the bottom. It also includes six studies using measures of mobility not confined to movement up from the bottom; these find either no change or rising mobility. In contrast, only two papers find a fall in mobility, each using non-directional measures. Notably, one of them shows an uptick in the mobility of the most recent two birth cohorts it examined, leaving in doubt the question of whether the longer-term decline it found would have persisted had the authors had more recent data. The other study finds somewhat mixed evidence, depending on the data source and whether children of single parents are included.

Of course this doesn’t mean mobility is “high enough.”  Scott also will be publishing a more detailed explication of these results with Brookings.


To examine subtle changes over time, you need an All Else Being Equal sample, such as just non-Hispanic whites, as Charles Murray uses in his upcoming book "Coming Apart."

And then, you still have to account for the impact of demographic change. For example, say your Dad was at the 15th percentile, but then the country lets in so many even poorer foreigners than your family, that you, even though you are just like your dad, are now at the 25th percentile. Upward mobility! Or maybe not ...

In general, all these debates over inequality and mobility need to deal with demographic change, but, as far as I can tell, most of them try to ignore the elephant in the room.

Wouldn't that depend on what you take "just like your dad" to mean? Make as much money? Make as much inflation adjusted money? etc. What's your definition?

Presuming you mean "just like your dad" to mean something to the effect of earning at a comparable purchasing power/inflation-adjusted income, wouldn't it require a significant decline in per capita GDP/income in order for that same purchasing power to slide up along the distribution scale?

Yes, I was going to make the same observation, and since we don't see that...

In fact, with today's poverty line about where 1950s average incomes lay, it's pretty amazing how much productivity gain our foreign immigrants gain by crossing our border.

(Or, at the very least, a large drop in incomes at the bottom of the distribution, offset by an increase at the top to keep the overall GDP figures steady or growing.)

Steve, the standard measure of mobility is the correlation between father/son earning at different points in the son's career (as in the linked Levine and Mazumder paper). If this correlation in earnings across generations falls over, we say mobility rises. Sure changes in the economy such as increased immigration, trade, or technological change could have an arguably stronger effect on the son's earnings...but the relationship is not mechanical in these studies.

Thanks Claudia for the description of the best methodology.

So, what you are also saying is that the variable (dad's income), if less correlated with son's income, declines in significance, then the effect on son's mobility lessens. But, that is not the same as saying that upward mobility could have, for example, a long recession, increased immigration, greater outsourcing, decline in trade related unemployment, etc. all effecting the son's mobility, along with everyone else's.

From the political perspective, I don't think politicians are choosing the variable father's income, but the total composite of all other variables affecting mobility, and even level of income, particularly those over which the government may have control..

Mobility as with inequality are relative concepts. The underlying trend in median income as discussed in TGS is a separate issue.

Honestly, can you define "intergenerational income mobility" precisely enough to make a derivative meaningful? It's already a second derivative (rate of change of a rate of change), and the underlying first derivative is gradual, strongly affected by economic cycles, and changes on timescales comparable to the time between generations.

Do economists have a polite way to say "you cannot possibly be measuring the effect you're claiming to measure"?

No way, and the climate sociologists got their back!


Economists do the best they can with the available data and analytical tools to answer questions of importance. (Same with a lot of other disciplines.)

Specifically with this case, what's wrong with the economist definition of mobility? Simply put it says: If my dad's income at age 40 is a bit above average among his generation, then, on average, should my income at age 40 to be a bit above average relative my generation...seems like a reasonable way to define mobility. (Other measures are possible.) Now economists (in the papers cited) measure this simple concept with conditional correlations (sometimes you want to exclude differences in education, sometimes not) in regressions with lots of statistical bells and whistles (to deal with a sundry measurement and conceptual issues).

Perfect no, but I don't think we have to apologize for trying.

My guess is that the income mobility story is going to depend on where on the income distribution you are looking. There's likely plenty of smart working class kids who manage to get to college and get some sort of professional job that moves them squarely into the upper middle class. And there's likely plenty of under-achieving upper middle class kids who regress to the mean. At the margins, the story is probably much different. Many of the poorest, born into crappy families and sent to crappy schools will stay that way. And many of the richest, with over-achieving yuppie parents who raise them as designer kids, won't fall too far down the income scale.

I think it actually is the opposite of that. Much more mean reversion at the very top and bottom.

Here you go.

Look at page 7.

Of course there are all sorts of tricky things going on in understanding what these data mean - the distance absolute distance between the 1st and 3rd quintiles is much larger than between the 3rd and 5th; a movement into an adjacent quintile could be due to a very small marginal change when comparing father to son.

In any case, those in the top are generally quite good at not regressing to the mean.

Can a society be upward income mobile without being downward income mobile too? I'm wondering about the redundancy of the epithet "upward" in all the income mobility discussions.

Assortative mating and heritability will also produce what appears to be limited upward mobility, but is just smarts begetting smarts.

This seems way too clever by half and counter intuitive to be taken so seriously.

I don't think we have any real intuition about mobility. I think we have intuitions about current wealth and income distributions and dangerously infer something about mobility.

Mobility in America would rise if we increased immigration from China mch less if we increased immigration from Mexico. So what is the meaning of this?

I wonder if they exclude immigrants? My wife's HHI has increased roughly 100-fold in her lifetime. Such people would presumably have an outsize impact...

It's not entirely clear to me what the people claiming that we don't have enough income mobility want to happen. Do they want children's income to be completely uncorrelated to their parents'?

Mobility isn't entirely (and possibly not even mostly) due to parents buying their children a head start. A big part of it is the cognitive and personality characteristics that children inherit from their parents, genetically and/or culturally. The only way you could actually get the correlation down to zero is by assigning incomes randomly, so the optimal correlation between parents' and children's incomes is clearly greater than zero.

It's also pretty obviously less than one. But there's no obvious way to know where it should be a priori, and this probably depends on the genetic and cultural diversity of the population.

It may be useful to think about trends in economic mobility along with trends in income inequality. People who believe that rising differences in income (inequality) reflects more rags to riches experiences (mobility) often have different views about government policies than people who think more inequality is the rich getting richer.

@GiT - Thanks for the link..........

There are a number of barriers to upward mobility starting with inheritance and income taxes and the credentialism spawned by publicly subsidized education. But there are also barriers to social mobility by the US's increasing resort to banana-republic policies by which the rich are protected from becoming poor. Thus, over-leveraged fools who paid too much are rescued at the expense of prudent savers and capital remains tied up.

The appropriate response to the 2008 crisis should have been to allow the de-leveraging to proceed. GM with its legacy costs and junk bonds should have been wiped out, along with AIG, Goldman Sachs and all the cash-rich fools buying McMansions in the suburbs and condo's wherever a developer could find a free patch of dirt. A lot of wealth did in fact get redistributed before government stepped in with TARP and other interventions. Government policies that make decent housing increasingly scarce, such as immigration, zoning, Section 8, etc., also erect high barriers to social mobility.

I have not taken the time to survey the literature on intergenerational *wealth* mobility, which is arguable a more important long-term indicator and incentive than income mobility. The data, I think, is scarce on the ground, but would be interested in hearing of any cogent and convincing work that's been done.

Here is part of the abstract from one of the studies that Winship references when he says "either that mobility has increased over the long run or that it has changed little in either direction. That includes both studies":

"this paper shows that the transmission of high-income status significantly increased while the transmission of low-income status remained stable. These results suggest that it has now become easier for high-income sons to maintain their economically advantaged status than in the past. In contrast, low-income sons’ chances of escaping from their economic disadvantage have not increased to the same extent."

Doesn't "increased transmission" mean decreased mobility?

Comments for this post are closed